Role of air-sea-land interactions on the marine boundary layer dynamics in the upwelling region off Central Chile

HAL Id: tel-02924557

https://tel.archives-ouvertes.fr/tel-02924557

Submitted on 28 Aug 2020

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

layer dynamics in the upwelling region off Central Chile

Orlando Astudillo

To cite this version:

Orlando Astudillo. Role of air-sea-land interactions on the marine boundary layer dynamics in the upwelling region off Central Chile. Ocean, Atmosphere. Université Paul Sabatier - Toulouse III, 2018.

English. �NNT : 2018TOU30360�. �tel-02924557�

TH `ESE TH `ESE

En vue de l’obtention du

DOCTORAT DE L’UNIVERSIT´E DE TOULOUSE

D´elivr´e par :l’Universit´e Toulouse 3 Paul Sabatier (UT3 Paul Sabatier)

Pr´esent´ee et soutenue le31/08/2018 par :

Orlando ASTUDILLO

Rˆole des interactions océan-atmosphère-continent sur la dynamique de la couche limite marine dans la région d’upwelling du Chili Central

Sylvain COQUILLAT ProfesseurJURY Pr´esident du Jury Boris DEWITTE Directeur de Recherche Directeur de thèse Marc MALLET Charg´e de Recherche Co-directeur de thèse Xavier CAPET Directeur de Recherche Rapporteur

Ren´e GARREAUD Professeur Rapporteur

Abderrahim BENTAMY Charg´e de Recherche Examinateur Christine PROVOST Directeur de Recherche Examinateur Aurore VOLDOIRE Charg´e de Recherche Examinateur

´Ecole doctorale et sp´ecialit´e :

SDU2E : Oc´ean, Atmosph`ere, Climat Unit´e de Recherche :

Laboratoire d’ ´Etudes en Géophysique et Océanographie Spatiales (UMR 5566) Directeur(s) de Th`ese :

Boris DEWITTE etMarc MALLET Rapporteurs :

Xavier CAPET et Ren´e GARREAUD

I would like to dedicate this thesis to:

Angélica, my dearest wife, who led me through this trip of learnings, efforts and discoveries with all her light of love, hope and support,

My beloved parents, Diogena and Orlando, who always give everything for his sons in countless ways,

My beloved sisters and brothers who encourage and support me always,

My nephews and nieces who always keep in my heart a big portion of tenderness and love.

Acknowledgements

I am very grateful to the many people who have support me along the development of this thesis.

First, I would like to acknowledge my supervisor, Dr. Boris Dewitte, for his help and advice during my period of research. I really appreciate all his contributions of scientific ideas and methods. I would also like to thank my co-supervisors, Dr. Marc Mallet and Dr. José Rutllant, for their valuable guidance during my studies. I am also thankful to my research collaborators Dr. Frédéric Frappart, Dr. Katerina Goubanova, Dr. Luis Bravo, Dr.

Serena Illig and Dr. Oscar Pizarro for his constructive comments and advices, which were of great help to improve this research. I am especially grateful to my former advisor Dr Melitta Fiebig and my colleague Dr. Marcel Ramos for introduce me into the numerical modeling of the atmosphere and ocean, encouraging me to pursue a Ph.D.

I would further like to thank to the Directors and Administrative members of the Laboratoire d’Études en Géophysique et Océanographie Spatiales (LEGOS), the Doctoral Department Sciences de l’Univers, de l’Environnement et de l’Espace (SDU2E) and the Université TOULOUSE III Paul Sabatier for allowing me to undertake this work.

Lastly, I would like to gratefully acknowledge the financial support of the Centro de Estu- dios Avanzados en Zonas Áridas (CEAZA). My thanks are also extended to Dr. Bernardo Broitman, Dr. Carlos Olavarría and Claudio Vazquez, Executive Directors and Corporate Manager of CEAZA, who endorsed my PhD. pursuit.

Les systèmes de courant de bords est (EBUS) sont les régions océaniques des latitudes tropicales à moyennes le long des côtes ouest des continents. Ils abritent des écosystèmes marins très productifs en raison de la circulation atmosphérique de surface dirigée vers l’équateur qui font remonter des eaux profondes froides (upwelling) enrichies en éléments nutritifs à l’origine de la vie marine le long de la côte. Si les processus océaniques fondamentaux de l’upwelling côtier sont bien connus (transport et pompage d’Ekman), la modélisation océanique des EBUS reste problématique en raison des difficultés pour prendre en compte de manière réaliste des phénomènes à fine échelle spatiale dans la zone de transition entre le littoral et l’océan du large.

Dans cette thèse, nous nous sommes concentrés sur le système d’upwelling dit de Humboldt (côtes du Pérou et du Chili) et sur l’influence des caractéris- tiques méso-échelles des vents près de la côte, en particulier la décroissance vers la cotes du vent (appelé “drop-off”) qui détermine l’importance relative des processus d’Ekman, et donc, la structure spatiale de la zone d’upwelling.

Une approche combinée basée sur l’analyse de données satellitaires et sur la modélisation régionale, océanique et atmosphérique, est utilisée pour étudier la sensibilité de la circulation océanique le long de la côte Chili central aux caractéristiques du drop-off.

Dans un premier temps, la circulation atmosphérique de surface moyenne à saisonnière le long du littoral du Pérou et du Chili est documentée pour la première fois à partir des données altimétriques de quatre missions satellites (ENVISAT, JASON1, JASON2 et SARAL). L’analyse révèle l’existence d’une réduction marquée de la vitesse du vent le long de la côte, bien que le taux de réduction varie en fonction de la latitude. Malgré la répétitivité relativement faible des satellites, nous montrons que les données altimétriques permettent néanmoins d’échantillonner le cycle saisonnier du drop-off. L’estimation de l’upwelling côtier à partir de ces données suggère que le pompage d’Ekman

vii

tend en moyenne à dominer par rapport au transport d’Ekman le long de la côte péruvienne, alors que le long de la côte chilienne, le transport d’Ekman est le processus dominant.

Dans un second temps, un modèle atmosphérique régional (WRF) à différentes résolutions horizontales (36 km, 12 km et 4 km) dans une configuration imbri- quée zoomée sur la région centrale du Chili a été développé afin de produire des champs atmosphériques présentant des caractéristiques différentes du drop-off.

Les solutions du modèle atmosphérique sont d’abord évaluées par rapport aux observations, indiquant un plus grand réalisme près de la côte que les réanalyses atmosphériques. Le rotationnel du vent cyclonique simulé le long de la côte associé au drop-off présente des échelles transversales comprises entre 8 et 45 km avec une variabilité latitudinale significative, en accord avec les vents altimétriques. Lorsque la résolution du modèle est augmentée, le drop-off est généralement d’autant plus confiné à la côte et le modèle indique une saisonnalité marquée avec un maximum d’intensité au printemps-automne. La contribution relative de la divergence côtière et du pompage d’Ekman présente une modula- tion latitudinale liée aux détails de l’orographie et de la ligne de côte.

Des expériences avec un modèle océanique régional (ROMS) sont ensuite réa- lisées pour estimer et comprendre l’influence du drop-off sur la circulation océanique et la dynamique de l’upwelling. Il est montré que la prise en compte d’un drop-off plus réaliste dans le forçage atmosphérique côtier induit à une ré- duction notoire de l’intensité du jet océanique côtier de surface, un sous-courant plus fort et une dérive d’Ekman cohérente avec les observations disponibles.

Les analyses de l’énergie cinétique turbulente et des flux de chaleur turbulent illustrent la réponse non linéaire de la dynamique d’upwelling à la représentation des caractéristiques du vent à méso-échelle. En particulier, alors que la prise en compte du vent dans le forçage atmosphérique du modèle océanique régional conduit à une réduction globale du biais froid côtier observé dans les simula- tions des modèles océaniques régionaux forcés par les produits atmosphériques couramment utilisés (i.e. Reanalyses globales et vents diffusiometriques), les résultats suggèrent également qu’il influence la dynamique de la couche limite de fond et de surface dans certaines régions et donc la position du front d’upwelling.

Nous discutons l’implication de nos résultats pour l’amélioreration des forçages atmosphériques dédiés à la modélisation océanique régionale dans les EBUS.

Eastern Boundary Upwelling Systems (EBUS) are the tropical to mid-latitudes oceanic regions along the west coast of the continents. They host very productive marine ecosystems owing to the mean equatorward low-level atmospheric circu- lation that uplifts cool subsurface nutrient-enriched waters that trigger marine life along the coast. While the fundamental oceanic processes behind such process are well known (i.e. Ekman transport and pumping), the oceanic modeling of the EBUS has remained problematic owing to difficulties in accounting realistically for phenomena at fine spatial scales in the transition zone between the littoral and the off-shore ocean.

In this thesis we have focused on the Peru-Chile Upwelling System (so-called Humboldt system) and on the influence of the cross-shore mesoscale features of the winds near the coast, particularly the shoreward wind drop-off, which determinate the relative importance of the Ekman processes, and thus, the spatial and temporal structure of the upwelling. A combined approach based on satellite data analysis and regional modeling, both oceanic and atmospheric, is used to investigate the sensitivity of the oceanic circulation along the coast of central Chile to the characteristics of the wind drop-off.

As a first step, the mean to seasonal near-shore surface atmospheric circula- tion along the coast of Peru and Chile is documented for the first time based on the altimeter data from four satellite missions (ENVISAT, JASON1, JASON2 and SARAL). The analysis reveals the existence of a marked shoreward re- duction in the wind speed all along the coast, although the reduction rate is latitudinally dependent. Despite the relatively weak repetitivity of the satellites, it is shown that the altimetric data are able to sample the seasonal cycle of the wind drop-off at some locations. The estimate of coastal upwelling from these data suggests that Ekman pumping tends on average to dominate with respect to Ekman transport over the Peruvian coast, whereas over the central-Chilean coast, the Ekman transport is the dominant process.

ix

In a second step, a regional atmospheric model (WRF) at different horizon- tal resolutions (36km, 12km and 4km) in a nested configuration zoomed over the central-Chile region was developed in order to produce atmospheric fields with different characteristics of the wind-stress curl (drop-off) along the coast. The atmospheric model solutions are first evaluated against the satellite observations, showing a much larger realism than atmospheric Reanalyses near the coast. In particular, the simulated cyclonic wind curl along the coast related to the wind drop-off exhibit length scales between 8 and 45 km with a significant latitudinal variability, which is in agreement with the altimetric winds. The higher model resolution, the more confined to the coast the wind drop-off, with the latter evidencing a marked seasonality with a maximum intensity in spring-fall and minimum in winter. The relative contribution of the coastal divergence and Ek- man pumping exhibits a latitudinal modulation linked to details in the orography and coastlines.

Experiments with a regional oceanic model (ROMS) are then carried out to estimate the dynamical impacts of the representation of the drop-off in the at- mospheric forcing of the ocean model. It is shown that the consideration of a realistic wind drop-off in the coastal atmospheric forcing induces a notorious reduction in the oceanic coastal jet intensity, a stronger poleward undercurrent and a coherent offshore Ekman drift in agreement with available observations.

Additionally, the analyses of the Eddy Kinetic Energy and eddy heat flux in the various sensitivity experiments illustrate the non-linear response of the upwelling dynamics to the representation of the mesoscale wind features. In particular, while the consideration of the wind drop-off in the atmospheric forcing of the regional oceanic model yields an overall reduction of the coastal cold bias found in the regional oceanic model simulations forced by commonly used available atmospheric products, the results also suggest that it can influence the surface and bottom boundary layers in some regions and thus the position of the up- welling front. Implications of our results for the improvement of regional oceanic modeling in EBUS are discussed.

Contents

1 Introduction 1

1.1 Preamble . . . 1

1.2 The Eastern Boundary Upwelling Systems . . . 2

1.2.1 The Carbon Cycle and Oceanic Carbon Pumps . . . 4

1.2.2 The Role of Upwelling in the Carbon Cycle . . . 6

1.2.3 The Oxygen Minimum Zone . . . 7

1.2.4 Eastern Boundary Upwelling System Dynamics . . . 9

1.3 Scientific Focus, State of the Art, Motivations and Objectives . . . 20

1.3.1 Climate Model Biases in the EBUS, Causes and Impacts . . . 22

1.3.2 Coupled Climate Systems of the Southeast Pacific . . . 24

1.3.3 Atmospheric Circulation, Coastal Wind Variability . . . 26

1.3.4 Response of the PCUS to Coastal Mesoscale Wind Structure . . . . 28

1.3.5 Impact of the Wind Stress Forcing on Coastal Circulation . . . 29

1.3.6 The Impact of Mesoscale Wind Patterns on Biological Productivity 32 1.3.7 Mechanism of Air-Sea-Land Interaction . . . 33

1.3.8 PCUS Dynamics Under Climate Change . . . 35

1.3.9 Scientific Objectives and Methodology . . . 37

1.4 Introduction (Français) . . . 39

2 The Mesoscale Atmospheric Circulation Along the Coast of Peru and Chile 41 2.1 Introduction . . . 41

2.2 Surface Winds Off Peru-Chile: Observing Closer to the Coast From Radar Altimetry: An Observational Study . . . 41

3 Seasonal Dynamics Off the Central Chile Upwelling System 60 3.1 Introduction . . . 60

3.2 Seasonal Variability of the Ekman Dynamics in the Upwelling System Off Central-Northern Chile (⇠30 S): a Modeling Study . . . 61

4 Response of the Central Chile Upwelling System to Coastal Wind Drop-Off 79 4.1 Introduction . . . 79 4.2 Sensitivity of the Near-Shore Oceanic Circulation Off Central Chile to

Coastal Wind Profiles Characteristics . . . 79

5 General Conclusions and Future Work 113

5.1 Conclusions et Perspectives (Français) . . . 119

List of Abbreviations 124

Bibliography 127

Chapter 1 Introduction

1.1 Preamble

Eastern Boundary Upwelling Systems (EBUS) are characterized by the phenomenon of rising cold waters under the action of winds parallel to the coast blowing towards the equator. These regions also correspond to the zones of subsidence of dry air masses associated with the descending branches of Hadley-Walker circulation. These atmospheric and oceanographic characteristics of the circulation make it the site of specific air-sea interactions with in partic- ular the formation of a low cloud cover (stratocumulus type) which reflects the solar flux and amplifies the cooling effect of upwelling. Currently the global models are failing to simulate circulation in these regions with the most important warm bias of the tropical belt.

This is particularly the case for the Humboldt Current System (HCS; Peru-Chile coasts), the most productive upwelling region in the world in terms of fisheries. The source of the biases in global models remains poorly known, although it is partly related to the generally too low resolution of these models to realistically simulate the upwelling phenomenon on the one hand and the structure of near-shore winds on the other hand. The latter are in fact characterized by a decrease from the ocean to the coast in a coastal strip of the order of a few dozen kilometers. This phenomenon, called "drop-off" conditions the dynamics of upwelling through the Ekman pumping process. In return, upwelling favors a stable atmo- spheric boundary layer and therefore low-level winds are decoupled from the winds aloft.

This process of interaction between the ocean and the atmosphere on a regional scale is still poorly understood and documented. However, it could be a key element in understanding the dynamics of these regions, which are highly sensitive to changes in global variability.

In order to better understand the dynamics of the Peru-Chile Upwelling System (PCUS), the

study area of this thesis, as well to introduce the mechanisms of air-sea-land interactions in PCUS and its implications on upwelling dynamics, this first chapter will present the main aspects of the EBUS, including the significant role that the coastal upwelling plays in the cycles of carbon and biological productivity. Following, we will describe the main processes that affect the EBUS mean circulation and upwelling variability. Finally, this introductory chapter concludes with the presentation of the scientific focus and motivations of the thesis, along with detailing the objectives and methodological approach.

1.2 The Eastern Boundary Upwelling Systems

Eastern Boundary Upwelling Systems (EBUS) are among the most productive marine ecosystems of the world oceans. While they cover only a tiny fraction of the ocean surface (⇠1%), they are responsible for most marine biomass production and account for ⇠20%

of the world’s fisheries (Fréon et al., 2009; Narayan et al., 2010). The four main EBUSs, California, Canary/Iberian, Benguela and Humboldt Currents Systems (HCS) are narrow regions of the coastal ocean that extend latitudinally over several thousands of kilometers and longitudinally to beyond the continental shelves whose widths ranging from 10 to 200 km. They are located, on the one hand, in the eastern part of the oceans, and on the other hand, on each side of the Equator from 10-20 in latitude, which are outlined in white in Figure 1.1a. In these regions, intense equatorward, alongshore winds combined with the earth’s rotation cause offshore surface Ekman transport, subsurface onshore flow and upwelling of water that is rich in nutrients, low in temperature (sometimes 10 C below the average of normal temperatures at these latitudes), depleted oxigen, and pH. The arrival of nutrients in the surface waters illuminated by the sun (euphotic zone) triggers an exceptionally biological productivity (Carr, 2001; Strub et al., 2013), as illustrated in Figure 1.1b by the high chlorophyll concentrations, a typical proxy for the amount of photosynthetic plankton, or phytoplankton (floating marine plants), present in the ocean. The numerous variety of marine life includes low trophic organisms like bacteria and photosynthetic plants (e.g., sea grasses, seaweeds, and phytoplankton), and larger animals (e.g., zooplankton, crustaceans, fish, birds and mammals), that interact through complex food webs.The dynamics of these interactions are based on the uptake of simple inorganic chemicals and their conversion to more complex organic material by bacteria and phytoplankton, these organisms are known as autotrophs because they produce their own organic matter. Higher on the food web they sustain heterotrophic organisms that require pre-manufactured organic matter (Kämpf and Chapman, 2016).

1.2 The Eastern Boundary Upwelling Systems 3

(a)Global mean sea surface temperature (in C - left panel), with the eastern boundary upwelling systems(EBUS) characterized by cold SST indicated in white boxes, and a zoom over the Humboldt current system (in C - right panel). From MODIS Aqua (2002-2017).

(b)Global mean surface chlorophyll a concentration (in mg m ³- left panel). From MODIS Aqua (2002-2017).

Figure 1.1

1.2.1 The Carbon Cycle and Oceanic Carbon Pumps

The basis for these marine ecosystems is the light-induced photosynthesis carried out by phytoplankton and phototropic bacteria confined to the upper 50 - 100 m of the water column.

Through carbon fixation there is conversion of inorganic to organic carbon, which allows marine organisms to grow and reproduce while producing oxygen as a by-product. During photosynthesis, phytoplankton fixes the dissolved inorganic carbon and assimilates nutrients (nitrates, phosphates, silicates, iron, etc.) that are available in the euphotic layer. When the organisms die, part of the organic matter already synthesized is degraded (or remineralized) by bacteria carrying out ammonification then nitrification in the euphotic layer, and the other part is exported to the deep layers where is remineralized or stored in sediments. The rate of carbon fixation is strongly controlled by the availability of nutrients through out dynamical processes. Nutrients are supplied to the euphotic zone via upwelling, vertical mixing, conti- nental runoff of sediment-laden waters via rivers or groundwater seepage, and to some extent by atmospheric dust deposition. Iron is the main limiting nutrient for the upwelling regions of the Humboldt Current (Hutchins et al., 2002); and the California Current (Chase et al., 2007; Hutchins et al., 1998).

The global carbon cycle is the interaction of both biogeochemistry and physics within the ocean; it plays a major role in regulating climate by controlling the amount of the green- house gas CO2in the atmosphere. From the start of the industrial revolution in the mid-18th century, the atmospheric carbon budget has been substantially disturbed through human activities, such as fossil fuel combustion and cement manufacture, so that the pre-industrial atmospheric CO2concentration of about 270 ppm now exceeds 403 ppm and is continuing to increase. Roughly 50 % of the CO2produced by human activities is taken up by the ocean, the remainder staying in the atmosphere where it contributes to global warming (Kämpf and Chapman, 2016).

The oceanic carbon cycle can be described by different carbon pumps (See Fig. 1.2), each describing specific mechanisms that transfer carbon dioxide from the upper to the deep ocean or vice versa. These pumps are thesolubility pumpand thebiological pump. Phytoplankton is the engine of thebiological pump(responsible for 80 % of total carbon fixation in the ocean) that helps maintain a steep gradient of CO2between the atmosphere and deep ocean.

The biologic pump comprises

a) the organic carbon pump, associated with primary production in the euphotic zone and remineralization of detritus at depths.

1.2 The Eastern Boundary Upwelling Systems 5

Figure 1.2.Sketch of the basic oceanic carbon cycle: on the left hand side, thebiological pump, which is a collective property of a complex phytoplankton-based food web; on the right hand side the solubility pump, which is driven by chemical and physical processes, both mechanisms transfer carbon dioxide from the upper to the deep ocean or vice versa, maintaining a sharp gradient of CO2 between the atmosphere and the deep ocean. From Chisholm (2000).

b) the calcium carbonate counter pump, associated with skeleton and shell formation in the surface ocean and the dissolution of calcareous particles at depth.

The solubility pump is responsible for about 20 % of the vertical gradient in dissolved inorganic carbon in the ocean. Thesolubility pumpis based on:

a) the temperature dependency of the solubility of carbon dioxide in seawater; i.e. under the same atmospheric conditions, cold water can dissolve more CO2than warm water before it reaches an equilibrium with the atmosphere.

b) oceanic flows that either export surface water to the ocean interior (called oceanic subduc- tion) or bring deeper cold water back to the sea surface (upwelling).

The deep circulation of the oceans driven by open-ocean convection in sub-polar regions of the North Atlantic Ocean is generally known as one of the branches of thesolubility pump.

The other branch is the reverse process of upwelling in which CO2enriched deeper water is returned to the sea surface (Kämpf and Chapman, 2016; Sarmiento et al., 1995).

EBUS have been classified as Large Marine Ecosystems (LMEs) in the first productiv- ity category (>300 g C/m²/yr), according to their annual Gross Primary Productivity (GPP), that is, the rate of conversion of CO2to organic carbon per unit surface area by autotrophic organisms (Sherman and Hempel, 2008), while respiration refers to the energy-yielding oxidation of organic carbon back to carbon dioxide. The basic equation is the same in both cases, although the two mechanisms operate in reverse. The chemical reactions is:

Energy+nutrients+6CO2+6H2O$C₆H₁₂O₆+6O2 (1.1) where CO2is carbon dioxide, H2O is the water molecule, energy comes from solar radiation, nutrients are the nutritive salts, C6H12O6is a sugar molecule and O2is the oxygen molecule.

Hence, primary production results in a decrease in carbon dioxide with production of oxygen, while respiration uses up the oxygen to break down organic matter. The process of respiration occurs continuously, while primary production in the surface ocean waters can only take place during daylight (Kämpf and Chapman, 2016).

1.2.2 The Role of Upwelling in the Carbon Cycle

Coastal upwelling regimes associated with eastern boundary currents are the "engine" of GPP and marine productivity in the ocean. Indeed, they play a key role in the microbially mediated cycling of marine nutrients. These systems are characterized by strong natural variations in carbon dioxide concentrations, pH, nutrient levels and sea surface temperatures on both seasonal and interannual timescales (Capone and Hutchins, 2013).

From a biological point of view, EBUS would represent carbon sinks since in the long- term the high CO2 fixation by phyto-plankton production exceeds plankton community respiration. However, in spite of their high productivity, upwelling systems can facilitate an outgassing of CO2into the atmosphere when atmospheric heating reduces the solubility of CO2(Paulmier and Ruiz-Pino, 2009). The production of other greenhouse gases, like nitrogen dioxide, methane and other volatile gases, may potentially trigger further acidification of

1.2 The Eastern Boundary Upwelling Systems 7 EBUS (Checkley and Barth, 2009). Certain phytoplankton groups are also known to con- tribute to increased fluxes of dimethyl-sulphide (Franklin et al., 2009), a trace gas involved in the global biogeochemical cycling of sulphur, which influences climate by inducing aerosol and cloud formation in the atmosphere. On the other hand, the high biological productivity of EBUS generates a significant amount of organic detritus (dead organic matter), this organic matter is degraded in the water column and/or at the sediment by bacterial activity. During the remineralization process, a large amount of O2is consumed, thus explaining the reduction of the oxygen concentration below the mixing layer and the formation of Oxygen Minimum Zones (OMZ), see Fig. 1.3.

Figure 1.3. Schematic vertical profile of water column processes showing the well oxy- genated euphotic zone, oxycline/upper nutricline region, suboxic zone and anoxic layer.

OM=organic matter. From Peña et al. (2010).

1.2.3 The Oxygen Minimum Zone

These are regions in the ocean tightly coupled to the EBUS (See Fig. 1.4), they are character- ized by extremely low concentrations of Dissolved Oxygen (DO) in the water column (DO

< 60 µM), and may frequently reach values of DO lower than 40 µM, in the suboxia range (Naqvi et al., 2010), or even lower levels may also be found in certain areas (Schunck et al., 2013; Ulloa et al., 2012).

Figure 1.4. Global Oxygen Minimum Zones, including (a) Upper depth (in meters) of intermediate water hypoxia ([O2]<1.4 ml L ¹) and(b)thickness (in meters) of intermediate water hypoxia ([O2]<1.4 ml L ¹). The geospatial distributions of severely hypoxic [O2] minimums (of [O2] = 0.5 ml L ¹ and [O2] = 0.2 ml L ¹) are depicted on both panels as white lines. Data from World Ocean Atlas. From Moffitt et al. (2015).

OMZs form at shelf and upper slope depths, and are considered to be unique biologi- cal, geochemical and evolutionary environments, analogous to cold seep or deep-sea vent environments (Moffitt et al., 2015). As continental margin ecosystems transition from well oxygenated surface waters to the hypoxic core of the OMZ ([O2] = 0.5–0.1 ml L ¹), faunal diversity, trophic structures and physiological strategies change (Levin et al., 1991). OMZ oxygenation gradients produce successional biological zonation and are fundamental habitat barriers for benthic and pelagic organisms (Moffitt et al., 2015). Additionally, the biogeo- chemical cycles that take place at extremely low DO concentrations are involved in the local

1.2 The Eastern Boundary Upwelling Systems 9 production of climatically-active gases, such as CO2(Paulmier et al., 2006) and N2O (Kock et al., 2016), which are then outgassed to the atmosphere. In this sense, the OMZ has an impact on both the local ecosystems and on the global climate.

For all these reasons, several international programme such as the Intergovernmental Panel on Climate Change (IPCC), the Integrated Marine Biogeochemistry and Ecosystem Research (IMBeR), the Surface Ocean - Lower Atmosphere Study (SOLAS), the Intergovernmental Oceanographic Commission of UNESCO (IOC-UNESCO) and the United Nations Envi- ronment Programme (UNEP) have labeled EBUS a priority study area for the past ten years.

1.2.4 Eastern Boundary Upwelling System Dynamics

Ekman Dynamics

The physical mechanisms that explain the occurrence of coastal upwelling in the EBUS have been known for a long time. Vagn Walfrid Ekman, who studied the ocean response to wind forcing according to the observations of his predecessor the Norwegian scientist and explorer Fridtjof Nansen founded the principles of the upwelling theory in 1905. In the 1890s, Nansen led an expedition across the Arctic ice. His specially designed vessel, the Fram, was allowed to freeze into the ice and drift with it for over a year. During this period, Nansen observed that ice movements in response to wind were not parallel to the wind, but at an angle of 20–40 to the right of it. W. Ekman, a PhD. student in that period, developed his theory of wind-driven currents in order to explain this observation.

First to present the Ekman’s solution to wind-driven current we will present the essentials of surface circulation:

Consider quasi-steady, large-scale motions in the atmosphere or the ocean, away from boundaries. For these flows an excellent approximation for the horizontal equilibrium is a balance between the Coriolis force and the pressure gradient:

f v= 1 r

∂p

∂x, and f u= 1 r

∂p

∂y. (1.2)

Here we have neglected the nonlinear acceleration terms, which are of order U²/L, in comparison to the Coriolis force ⇠fU (U is the horizontal velocity scale, and L is the horizontal length scale). The ratio of the nonlinear term to the Coriolis term is called the

Rossby number:

Rossby number=Nonlinear acceleration

Coriolis force ⇠ U²/L fU = U

f L =Ro. (1.3) For a typical atmospheric value ofU⇠10ms ¹,f ⇠10 ⁴s ¹andL⇠1000km, the Rossby number turns out to be 0.1. The Rossby number is even smaller for many flows in the ocean (Ro⇠10 ³), so that the neglect of nonlinear terms is justified for many flows.

The balance of forces represented by 1.2, in which the horizontal pressure gradients are balanced by Coriolis forces, is called ageostrophic balance. In such system the velocity distribution can be determined from a measured distribution of the pressure field. The geostrophic equilibrium breaks down near the equator (within a latitude belt of±3 ), where f becomes small. It also breaks down if the frictional effects or unsteadiness become important (Cohen et al., 2004).

Ekman Transport

Like the atmosphere, the ocean is a continuous medium and satisfies conservation of mass, momentum and energy. Unlike the atmosphere, the ocean is very nearly incompressible.

∂u

∂x+∂v

∂y+∂w

∂z = 0 continuity equation, incompressible, (1.4) du

dt = 1 r

∂p

∂x+f v+Fx⇡0 u-momentum equation, geostrophic+friction, (1.5) dv

dt = 1 r

∂p

∂y f u+Fy⇡0 v-momentum equation, geostrophic+friction, (1.6) dw

dt = 1 r

∂p

∂z g⇡0 z-momentum equation, geostrophic, (1.7) whereFxandFystand for the vector components of friction per unit mass in the fluid,zis (close to) zero at the surface and decrease downward. If there are no accelerations (i.e. steady state and zero, or at least negligible advection accelerations), then ^du_dt =^dv_dt =0 and we are left with a balance of three forces on unit mass and thehorizontalequations of motion become:

f v+Fx 1 r

∂p

∂x = 0, (1.8)

f u+Fy 1 r

∂p

∂x = 0, (1.9)

1.2 The Eastern Boundary Upwelling Systems 11 i.e. Coriolis + Friction + Pressure = 0

as show schematically in Figure. 1.5.

Figure 1.5.Three forces in equilibrium on a water parcel. From Pond and Pickard (1983).

For the frictional forcesFxandFy, Newton’s Law of Friction states that in a fluid, the friction stresst, which is the force per unit area on a plane parallel to the flow, is given by

t =µ∂u

∂z = rv∂u

∂z. (1.10)

The quantity µ is the coefficient of (molecular)dynamicviscosity, whilev= _r^µ is the coeffi- cient of (molecular)kinematic viscosity. In the ocean where the motion is generally turbulent, the effective value of kinematic viscosity is the eddy viscosity with valuesAzof up to 10 ¹ m ²s ¹ for vertical shear (e.g. ^∂u_∂z). Then the eddy friction stress t=rAz∂u

∂z express the force of one layer of fluid on anareaof its neighbour above or below, but for substitution in the equation of motion we need an expression for the force on amassof fluid:

the force per unit mass = 1 r

∂t

∂z = 1 r

∂

∂z

✓

rAz∂u

∂z

◆

. (1.11)

The form of expression 1.11 whereA_z is assumed to be constant and consistent with the Boussinesq approximation becomes:

friction force per unit mass =Az∂²u

∂z². (1.12)

To simplify the problem, Ekman (1908) assumed the water to be homogeneous and that there was no slope at the surface, so that the pressure terms would be zero. He also assumed an infinite ocean to avoid the complications associated with the lateral friction at the boundaries.

Under these conditions, the horizontal equations of motion reduces to a balance between the Coriolis force and wind friction at the ocean surface (the stress acts along the wind direction):

r1∂p

∂x = f v+_r^1∂t_∂z^x = f v+Az∂²u

∂z² = 0

r1∂p

∂y = f u+_r^1∂t_∂z^y = f u+Az∂²v

∂z² = 0 9=

;the Ekman equations (1.13) i.e. Coriolis+Friction = 0, as in Figure. 1.6a.

The solution to Ekman’s equations are (in the case of a southerly wind):

u_E = ±V₀cos(^p₄+_D^p_Ez)exp(_D^p_Ez), v_E = ±V₀sin(^p₄+_D^p_Ez)exp(_D^p_Ez),

)

(+(-) for northern(southern) hemisphere) (1.14) where

V₀= ⁽

p2ptyh)

DEr|f| is the total Ekman surface current,

t_yh is the magnitude of the wind stress on the sea surface (approximately proportional to the wind speed squared and acting in the direction of the wind),

|f|is the magnitude of f,

DE =pq_2A

z

|f| is theEkman depthordepth of frictional influence.

The surface current is rotated 45 to the right (northern hemisphere) or left (southern hemi- sphere) of the wind. Further interpretation can be found in Pond and Pickard (1983).

The wind-driven Ekman current has its maximum speed at the surface and the speed decreases with depth. Because the strongest currents are to the right (or left) of the wind direction in the Northern (Southern) Hemispheres, it is easy to appreciate that the net transport will be to the right (or left) of the wind direction (Pond and Pickard, 1983). The basic form of the equations for horizontal motions (equations 1.13 ) in the absence of any pressure gradient is

rf v+∂t_x

∂z = 0 rf u+∂t_y

∂z = 0 (1.15)

which we can write as

rf vdz = dtx rf udz = dty (1.16)

1.2 The Eastern Boundary Upwelling Systems 13 Nowrvdzis the mass flowing per second in the y-direction through a vertical area of depth dzand width one metre in the x-direction, and^R_z⁰rvdzwill be the total mass flowing in the y-direction from the level z to the surface for this strip 1 m wide, while^R_z⁰rudzwill be the total mass transport per unit width in the x-direction. If we use the symbolsM_yE andM_xE to represent the Ekman mass transport in the y- and x-direction respectively, then

f M_yE = f^R0_2DErvdz= ^R0_2DEdtx= t_xh f MxE = f^R0_2DErudz= ^R0_2DEdty=t_yh

)

Ekman mass transport (1.17)

Now(tx) _2DE and (ty) _2DE will be essentially zero because the velocity below the wind- driven layer is substantially zero and therefore there can be no shear and therefore no friction (Pond and Pickard, 1983).

Figure 1.6. Wind-driven currents from Ekman analysis (a) net frictional stress balances Coriolis force with surface currentV0perpendicular to both; (b) wind in y-direction, surface velocityV0and components; (c) perspective view showing velocity decreasing and rotating clockwise with increase in depth; (d) plan view of velocities at equal depth intervals, and the

“Ekman spiral” (all for northern hemisphere). From Pond and Pickard (1983).

Ekman Pumping

Because of conservation of mass, horizontally divergent surface Ekman transport must be balanced by vertical motion (upwelling or downwelling). From the integral over the Ekman layer of the continuity equation, we have

r^{Z 0}

DE

✓∂u

∂x+∂v

∂y+∂w

∂z

◆

dz = 0 (1.18)

insertingM_xE andM_yE (i.e. equations 1.17 integrated from bottom of Ekman layer to surface)

∂

∂x Z ₀

DE

rudz+ ∂

∂y Z ₀

DE

rvdz = r(w(0) w( D_E)) (1.19) or equivalently

∂M_xE

∂x +∂M_yE

∂y =rwE (1.20)

equivalent to

—_H·ME=rw_E (1.21)

where—H·= [i(_∂^∂_x) +j(_∂y^∂ )]is the horizontal divergence operator andMis the vector mass transport (Pond and Pickard, 1983). Using 1.17, we find that Ekman pumping is proportional to the curl of the wind stress:

w_E= 1

rbk·—⇥

✓!t f

◆

(1.22) Sverdrup Balance

The interior flow, i.e. below the Ekman layer, of the (non-equatorial) oceans can be described in terms of its meridional circulation. In the subtropical gyres, the interior flow is toward the equator in both the Northern and Southern Hemispheres. In the subpolar gyres, the interior flow is poleward in both hemispheres. These interior flow directions can be understood through a potential (i.e. relative + planetary) vorticity argument introduced by Sverdrup (1947), so called the Sverdrup ‘balance”. As a reminder, for a frictionless, barotropic fluid (the interior flow), the potential vorticity of a fluid element of depth H is conserved:

✓z+f H

◆

= constant. (1.23)

The fluid element’s absolute vorticityz+f can respond to vertical stretching of the element by increasingz (ciclonic(anticiclonic) rotation for the northern(southern) hemisphere) or by migrating poleward (increased f).

1.2 The Eastern Boundary Upwelling Systems 15 Interior ocean is in geostrophic balance, then if we cross-differentiate 1.2 and subtract them, to eliminate the pressure terms, we get:

f

✓∂u x +∂v

y

◆ +d f

dyv = 0, (1.24)

By the continuity equation, and withb =^{d f}_dy = ^2Wcosf_R : bv= f∂w

∂z. (1.25)

This important equation states that water column stretching in the presence of rotation is balanced by a change in latitude (See Fig. 1.7). In 1.25 the vertical velocity wis due to Ekman pumping. Integrating this vertically through the interior ocean and from 1.21 and 1.22:

bM_y=

Z brvdz= frwE= fbk·—⇥

✓!t f

◆

⇡bk·—⇥ !t (1.26) or

My= 1

bbk·—⇥ !t (1.27)

This expression is known as the Sverdrup Balance and it equates the curl of the surface wind stress to the north-south transport over the water column integrated to the depth of no motion.

Figure 1.7.Sverdrup balance circulation (Northern Hemisphere). Westerly and trade winds force Ekman transport, creating Ekman pumping and suction and hence Sverdrup transport.

From Talley (2011).

Coastal Upwelling

Coastal upwelling owes its existence to the presence of both the coast as an impermeable lateral boundary and the wind blowing at the surface of relatively shallow water on the continental shelf. Under the action of the wind, the atmosphere exerts a tensile force (frictional force) on the ocean surface. The surface water is then drawn in the direction of the wind, but following a deflected trajectory to the right in the northern hemisphere and to the left in the southern hemisphere because of the Coriolis force linked to Earth’s rotation. The drift is offshore if the wind blows with the coast on its right (left) in the Southern (Northern) Hemisphere (Cushman-Roisin, 1994). The canonical structure of the oceanic circulation in coastal upwelling situations consists of a coast-parallel geostrophic current, called coastal jet, which is deflected seaward in an Ekman layer near the surface and shoreward in an Ekman layer near the bottom (See Fig. 1.8). The coast-parallel component of the wind stress induces offshore movement in the surface Ekman layer that operates to lower the coastal sea level by about⇠5–10 cm until a dynamical equilibrium is reached. This drop in sea level is sufficient to create a shoreward pressure-gradient force driving an inmediate geostrophic upwelling coastal jet of 20–50 cm s ¹in speed. In turn, the geostrophic flow becomes subject to frictional effects of the seafloor. This creates a shoreward flow in the bottom Ekman layer, which is about 5–25 m thick. This near-bottom flow is the final agent of the upwelling process as it moves near-bottom water shoreward and, as it hits the coast, upward into the euphotic zone (Kämpf and Chapman, 2016).

Wind Stress Curl Driven Upwelling

In addition to coastal upwelling, there is an additional key physical mechanism known as Ek- man pumping. This mechanisms refers to the vertical adjustment of the pycnocline associated with a spatially varying horizontal wind-stress field, known as wind-stress curl (See Fig. 1.9), which induces a divergence of horizontal Ekman transports in addition to boundary effects due to the existence of a coast. While the coastal upwelling process involves a cross-shelf transfer inherent with the dynamics of Ekman layers, Ekman pumping exclusively induces a vertical flow. Previous studies indicate that Ekman pumping contributes significantly (>25

%) to the vertical nutrient flux in major eastern boundary upwelling regions (Messié et al., 2009). A significant difference between the effects of wind stress and wind stress curl is that Ekman pumping caused by the latter takes place considerably farther offshore, often at the shelf edge, and can occur at faster rates in terms of vertical transport than Ekman transport

1.2 The Eastern Boundary Upwelling Systems 17 (Kraus and Businger, 1994). Thus, it is separated from the inshore Ekman transport and can provide a second input of nutrients to the system (Kämpf and Chapman, 2016).

Figure 1.8.The general dynamic structure of coastal upwelling. From Kämpf and Chapman (2016).

Figure 1.9.The general dynamic structure of the EBUS. Adapted from (Mohrholz et al., 2014).

Cross-Shelf Structure of Coastal Upwelling

As seen in the previous sections, Ekman transport can produce upwelling through flow divergence in the coastal area due to the coastal boundary and in the offshore area due to the wind curl driven vertical flow. The Ekman Coastal Upwelling Index (ECUI) derived by Bakun (1973), provides an estimation of the vertical velocities associated with the former upwelling based on both the Ekman transport and the cross-shore scale of the coastal upwelling cell structure (Lu) as follows:

ECUI=U_ek

L_u = t_alongshore

rf L_u (1.28)

whereU_ek is the Ekman transport, t_alongshore is the alongshore wind stress and r is the background density of sea water. Usually, for lack of better estimations,Lu is mistakenly taken as the internal Rossby radiusR⁰=q_g0h

f0 (withg⁰the reduced gravity andh₀the small Ekman upper layer thickness). This scale varies between 5 and 30 km depending on the stratification ratio, and describes the geostrophic adjustment scale of the pycnocline slope and not the cross-shore width of the upwelling cell. Indeed, Estrade et al. (2008) show, using an analytical model, that over shallow depths, coastal upwelling occurs in the frictional inner-shelf, where surface and bottom Ekmal layers overlap. This essential results reveal, on the one hand, how the inner-shelf geometry influences the coastal upwelling scale, and, on the other hand, how the cross-shore wind component drives the near-shore pressure gradient adjustment. The latter leads to an “migration” of the main upwelling cell with a separation from the coast driven by outcropping and homogenization of the water column where the surface and bottom boundary layers are fully merged (inner shelf zone) establishing a “kinematic barrier” to the the Ekman transport divergence, and, coastal incursion driven by a “boundary layers splitting” process caused by shoreward advection of the isopycnal uplift and stratification of the inner shelf. The former mechanism of upwelling separation from the coast could be weakened in case of severe coastal wind drop-off because frictional activity decreases in the near-shore region bringing the kinematic barrier and upwelling closer to the coast.

Overall, the results of Estrade et al. (2008) characterize the geography of the wind-driven upwelling structure. They provide an estimation of vertical velocity which can be used as coastal upwelling indices derived entirely from Ekman’s theory. For a local dephh, a constant topographic slopeSand Ekman depthDE (defined in equations 1.14), they show that 90%

of the Ekman transport upwells for _D^h_E between 1.25 and 0.5 (in the case of an alongshore wind), meaning that ^D_S^E (the ratio of Ekman depth to topographic slope) is the right scale

1.2 The Eastern Boundary Upwelling Systems 19 to estimate the cross-shore width of an upwelling cell, this result illustrates the frictional nature of coastal upwelling divergence. However, there is a lower limit to the inner-shelf scale when steep slopes are considered as horizontal friction becomes the dominant process in the creation of a frictional boundary layer. In this case,L_uis better scaled by the horizontal Ekman layer widthL_E =pq_2A

h

|f|.A_his the lateral eddy viscosity which is estimated in the range 10 100m²s ¹ in coastal upwelling regions (Marchesiello and Estrade, 2010) and depends upon submesoscale activity over the shelf (Capet et al., 2008). Figure 5 summarize the various physical processes at work as a function of depth.

Figure 1.10.Conceptual scheme of the mechanism of upwelling separation from the coast.

Adapted from (Estrade et al., 2008).

Poleward Undercurrents and Eddies

An additional dynamical feature of the EBUS is the presence of Poleward UnderCurrent (PUCs) over the shelf and slope, commonly found in midlatitude Eastern Boundary Current (EBC) systems (Fonseca, 1989; Hill et al., 1998; Neshyba et al., 2013). PUC can provide subsurface onshore flow to the upwelling systems where their properties may be significant in determining the biochemical response of the coastal environment to upwelling (phytplank- ton blooms and subsequent hypoxic decay) (Strub et al., 2013). These subsurface currents generally reach average velocities along the coast of⇠0.05 m/s to⇠0.2 m/s, distributed between depths of 150 and 300 m, as reported for the California Undercurrent and the Iberian

Peninsula (0.2 m/s). In the HCS, direct measurements of the PUC have been made off the coast of Peru and Chile. Based on 6 years of data at 30 S, Shaffer et al. (1999) reported an average velocity of 12.8 cm/s at 220 m, with semiannual variation and an intensified poleward flow in spring. The semi-annual variations of the PUC in the HCS are responses to the combination of local wind stress curl and disturbances caused by remote tropical forcing. This flow is also modulated interannually by Rossby waves forced by trapped waves propagating southward along the coast of South America, a physical mechanism that also partly explains the variability in the oxygen minimum zone off Chile (Aguirre et al., 2012;

Vergara et al., 2016).

On the other hand, upwelling jets are not smooth (laminar) flows. Like other frontal flows (e.g. western boundary currents) upwelling jets quickly become dynamically unstable on time scales of days to weeks and shed mesoscale eddies. In the four EBUS, these instabilities favor the generation of mesoscale eddies that are mainly formed near the coast and propagate mostly westward toward the interior of subtropical gyres (Chaigneau et al., 2009; Chelton et al., 2011; Morrow and Le Traon, 2012; Pegliasco et al., 2015). These structures are circular flow patterns in which the geostrophic flow surrounds a high-pressure (low-pressure) centre associated with an elevated (depressed) sea level. It is the low-pressure eddies (i.e. cyclones) that largely sustain upwelled water in their centres. Eddies in the coastal ocean can have diameters as small as 10–20 km, in contrast to the more commonly known open-ocean eddies that have diameters of up to 300 km. As a result of eddy shedding, the width of the upwelling zone generally increases along the coast in the direction of the upwelling jet. In addition, upwelling jets often form turbulent wakes in the lee of headlands. Fully developed eddy fields exhibit specific pathways, called filaments, along which upwelled water is advected offshore (See Fig. 1.11b). Filaments, which can be quasi-stationary or transient features, generally operate as an export mechanism of organic matter to the open ocean (Arístegui et al., 1997). Eddies also operate to disperse properties of the upwelling centre (e.g. heat anomalies, organic matter and zooplankton and fish larvae) offshore. These nonlinear mesoscale eddies trap water into their cores and act as transport and mixing mech- anism redistributing physical and bio-geochemical properties from the coastal regions to the open ocean (Barton and Arístegui, 2004; Dong et al., 2014; Logerwell and Smith, 2001;

Morales et al., 2012; Rubio et al., 2009). Along their paths, they can also modulate the biogeochemistry and ocean productivity (Correa-Ramirez et al., 2007; Gruber et al., 2011;

Mahadevan, 2014; Marchesiello and Estrade, 2007; Pegliasco et al., 2015; Stramma et al., 2013) and also impact the overlaying atmosphere interactions affecting heat fluxes at the

1.2 The Eastern Boundary Upwelling Systems 21 sea-air interface, winds, cloud cover and precipitations (Frenger et al., 2013; Mahadevan, 2014; Morrow and Le Traon, 2012; Pegliasco et al., 2015; Villas Bôas et al., 2015).

(a) (b)

Figure 1.11.Example of mesoscale structures including eddy field and upwelling filaments seen in satellite images, MODIS (Aqua), (a) Sea Surface Temperature (L3, Day, 8 Day, Thermal, 4 km) and (b) clorophyll-bconcentrations in the PCUS (28/02/2017). From NASA Worldview https://worldview.earthdata.nasa.gov.

1.3 Scientific Focus, State of the Art, Motivations and Ob- jectives

The oceanic region located along the west coasts of South America (so called Humboldt system) is now recognized as a key region for understanding the evolution of the climate in a warming world. This is primarily due to two main aspects, one related to the biogeochemical environment and the other one to the physical setting. This region is, first, embedded into the most extended and marked Oxygen Minimum Zone (OMZ) of the world (Paulmier and Ruiz-Pino, 2009) that results from the low ventilation of the oceanic circulation and that intervenes in the carbon and nitrogen cycles at global scale (Gruber et al., 2009). The OMZ is also favorable for the development of hypoxic events along the coast that can severely disrupt the rich ecosystem and so the marine resources that are crucial for regional economies.

Second, this region is involved in the climate system through up-scaling effects, the up- welled cold waters feeding back on the marine boundary layer and participating in the maintenance of low-level clouds under the subsidence inversion, the so-called Stratocumulus Cloud Deck (SCD), that influence the earth radiative budget (Bony and Dufresne, 2005).

These regions of low-level stratus tend to act as a thermostat for the ocean, blocking solar radiations at the ocean surface and, therefore, contribute significantly to Earth’s radiation balance (Mechoso et al., 2014).

1.3.1 Climate Model Biases in the EBUS, Causes and Impacts

Current generation Coupled General Circulation Model (CGCM) have the most severe SST biases in this region (see Fig. 1.12), and may frequently reach in excess of 5 (de Szoeke et al., 2010; Manganello and Huang, 2008; Richter, 2015; Yu and Mechoso, 1999). These warm bias are caused by several factors:

a) The bad representation of the stratocumulus deck, that implies a misrepresentation of the surface shortwave radiation that lead to an overestimation of the solar heat flux.

b) The underestimation of the mesoscale eddy activity that inhibit the offshore transport of cold waters due to the low spatial resolution of these global models.

c) The misrepresentation of the alongshore winds that impact the cooling associated with Ekman dynamics.

d) The normally sharp vertical temperature gradient separating the warm upper ocean layer from the deep ocean is too diffuse in the models

1.3 Scientific Focus, State of the Art, Motivations and Objectives 23

Figure 1.12. (a)Observed annual mean sea-surface temperature (SST) from the optimally interpolated (OI) SST data set.(b)Annual mean bias of the CMIP5²ensemble relative to OISST. The gray boxes denote the EBUS. From Richter (2015).

Such concern motivated in 2006 a specific international program (VOCALS/VAMOS, http://www.eol.ucar.edu/projects/vocals/) under the auspices of CLIVAR (Climate Vari- ability and Predictability of the ocean-atmosphere system) to address this issue over the offshore South-Eastern Pacific (SEP) sector (Wood et al., 2011). In preparation for VOCALS, a preliminary model assessment (PreVOCA) was conducted for October 2006 by operational forecast, regional, and global models, with a particular focus on the clouds and the Marine Boundary Layer (MBL) in the SEP (Wyant et al., 2010). Results in terms of large-scale dynamics (i.e. observed anticyclonic surface winds) were in agreement with observations but performed poorly on the representation of the stratocumulus with a significant dispersion within the models for the geographic patterns of mean cloud fraction with only a few models agreeing well with MODIS observations. Furthermore, most models also underestimate the MBL depth by several hundred meters (about one-half the observed values) in the eastern part of the study region (See Fig. 1.13). The shallow MBL in the PreVOCA models is usually

accompanied by the lack of clouds in the near-coastal region, but the nearshore surface wind field parallel to the coast is well reproduced in general.

Figure 1.13.Model boundary layer depth (m) compared with observations of boundary layer depth and cloud-top height. From Wyant et al. (2010).

Finally, in addition to the misrepresentation of local processes and/or ocean-atmophere inter- actions, global connections among regional biases or errors have been found. In fact, Wang et al. (2014) found in 22 climate models, participants in the Coupled Model Intercomparison Project phase 5 (CMIP5), that SST biases are commonly linked with the Atlantic Meridional Overturning Circulation (AMOC), which is characterized by the northward flow in the upper ocean and returning southward flow in the deep ocean. Furthermore, the EBUS seems to be particularly touched by the effect of these remote biasses, this has several outcomes and implications. First, the improvement of regional processes may not suffice for overall better model performance, because remote influences may override them. Second, a better understanding of these global teleconnections is necessary to improve the climate model performance.

1.3 Scientific Focus, State of the Art, Motivations and Objectives 25

Figure 1.14.Global SST bias and its relationship with the AMOC. (a), The annual-mean SST bias averaged in 22 climate models. The SST bias is calculated by the SST difference between the model SST and extended reconstructed SST. The dots denote where at least 18 of 22 models (82%) have the same sign in the SST bias. The rectangles represent the focused regions. (b), (c)Spatial maps of SST bias and the AMOC for the first inter-model SVD mode (accounting for 45% of total covariance). (d), Their corresponding coefficients. The x axis in (d)represents different models. From Wang et al. (2014).

1.3.2 Coupled Climate Systems of the Southeast Pacific

The VOCALS program permitted to gather a unique data set and to improve our knowledge of the processes of cloud formation in this region (Mechoso et al., 2014). It also shed light on the importance of mesoscale coastal dynamics in some particular regions along the coast in building up the regional atmospheric circulation. In late spring 2009 the Chilean Upwelling Experiment (CUpEx), a joint program to VOCALS, that took place in the near- shore region of (30 S), focused on the ocean-atmosphere interaction in a major upwelling centre off northern Chile (Garreaud et al., 2011). This experiment provided additional detailed information on coastal processes in a region that experiences the seasonal meridional migration of an atmospheric Low-Level Jet (Garreaud and Muñoz, 2005) having a significant impact on the upwelling seasonal variability (Aguirre et al., 2012; Renault et al., 2009). Overall the results gathered within VOCALS and CUpEx suggest an upscalling effect of the processes at fine scales (the scale of the upwelling cells) upon the regional climate variability, that is, processes of air-sea interactions taking place in a narrow coastal fringe (width of⇠100km) can feedback on the regional climate variability through their impact on the oceanic variability. The central Chile region is in fact located in the southern edge of the SCD of the SEP and the atmospheric Low-Level Jet (see Fig. 1.15) along the Chilean coast is often located at 30 S, particularly in winter and spring (Garreaud and Muñoz, 2005). Consistently, this region is recognized as one of the most active upwelling centres in Chile (Figueroa and Moffat, 2000), presumably in connection with local southerly wind maxima, and as a source of ocean kinetic energy ((see Fig. 1.16b) ) along the Chilean coast, especially during springtime (Hormazabal, 2004;

Rutllant and Montecino, 2002).

1.3.3 Atmospheric Circulation, Coastal Wind Variability

As mentioned above, the Low-Level Jet (LLJ) off central-Chile is a major feature of the atmospheric circulation along the western coast of South-America. It is forced synoptically by the passage of mid latitude migratory anticyclones farther south and results from the equilibrium between the alongshore pressure gradient and the turbulent friction in the MBL (Muñoz and Garreaud, 2005). Recent modelling studies (Rahn et al., 2011; Renault et al., 2012) indicates that the LLJ is associated to finer scale wind jets (⇠5km) along the coast, tightly linked to orography (see Fig. 1.15b). These mesoscale wind maxima associated are referred as coastal jets (CJs) to differentiate them from the broader, synoptic-scale LLJ. Due to their high variability, they can act as a driver of mixing near the coast eroding the uplifting of the isotherms and thus the upwelling effect at low frequencies. In addition, satellite data reveals areas of minimum low-level cloud frequencies located downstream of those points

1.3 Scientific Focus, State of the Art, Motivations and Objectives 27

Figure 1.15. (a)Spring-Summer (SONDJF) average of sea level pressure (contoured every 2 hPa) and 10-m wind vectors (arrows) off the Chilean coast. Data source: NCEP-NCAR reanalysis.(b)Spring-Summer average of surface wind speed derived from 4 yr of QuikSCAT observations. Color scale at right in m s 1. Note the near coastal jets off points Choros (Cho), Lengua de Vaca (LdV) and Lavapie (Lav). Adapted from Garreaud and Muñoz (2005).(c) Spring (SON) climatology of low cloud frequency derived from visible GOES imagery (pink is >80%; blue is less than 30%). From Garreaud et al. (2011).

and capes (Garreaud et al., 2011), indicative of topographically induced alongshore variability in the Atmospheric Boundary Layer (ABL) structure (see Fig. 1.15c).

Another feature is an overall wind reduction (not for the CJs) in a narrow coastal fringe of⇠30 km width. The latter is often referred to as the wind drop-off that is thought to be a main feature of the regional atmospheric circulation in upwelling regions (Capet et al., 2004). Several observational and modeling studies have indeed suggested that the mean wind stress in upwelling regions may be systematically reduced (See Fig 1.16a) within a narrow coastal strip (of⇠50km), relative to values farther offshore (Bane, 2005; Capet et al., 2004;

Perlin et al., 2007; Renault et al., 2012). For instance, Dorman et al. (2006) showed from observations that the summer-mean alongshore wind stress over the shelf off Bodega Bay (California) decreases from 0.14 N m ²at 25 km offshore to 0.04 N m ²at 2 km. During the CupEx experiments, there is also evidence that such cross-shore wind reduction (drop-off) is taking place off Coquimbo (30 S) (Garreaud et al., 2011).

Figure 1.16.Map illustrating the main features of the surface circulation off central Chile (spring 2008), as simulated by the regional atmospheric (WRF) and oceanic (ROMS) models:

(a) Surface wind magnitude/direction (m s ¹) and (b) Sea surface temperature ( C). In addition, acronyms indicate upwelling centers( PCho: Punta de Choros, PLdv: Punta Lengua de Vaca and PCur: Punta Curaumilla. From Astudillo et al. (2018), submitted, see Chapter 4.

1.3.4 Response of the PCUS to Coastal Mesoscale Wind Structure

The spatial and temporal structure of the upwelling is mainly driven by the coastal wind variability, in particular, cross-shore variations of the wind have received considerable attention, since the presence of a wind drop-off close to the shore tends to increase (decrease) upwelling through Ekman pumping (transport). Regional oceanic modelling studies show that upwelling response is highly sensitive to such a transition wind shape (Capet et al., 2004; Desbiolles et al., 2014, 2016; Jacox and Edwards, 2012). For instance, in a pionnered modelling study, Capet et al. (2004), carried out twin experiments for the California Current System (CalCS) that differ only in the cross-shore gradient of the nearshore wind, i.e. variable vs uniform drop-off, in a coastal strip 30-km wide. With this differentiated surface forcing, the SST near the coast for the smooth case is 2 C colder than the sharp case (see Fig. 1.17a), for a major upwelling event, revealing that nearshore wind drop-off diminishes upwelling;

i.e., the hypothesized compensation between nearshore Ekman transport and upward Ekman pumping does not fully occur. Additionally, they show the preponderant role of the coastal wind profile in determining the mean alongshore current structure (i.e. surface nearshore coastal jet and poleward undercurrent).

However, wind analyses do not represent adequately the mesoscale wind patterns (e.g.

COAMPS), and show uncertainties in the crosshore wind profile (see Fig. 1.17b), i.e the drop-off takes place over an increasingly small region as the model resolution increases, overestimating the wind drop-off magnitude which influence the nearshore circulation and

1.3 Scientific Focus, State of the Art, Motivations and Objectives 29

(a)SST in the Southern California Bight for a major upwelling event: observed(left)and modeled with a variable-smooth (middle) or uniform-sharp (right) crosshore gradient. From Capet et al.

(2004).

(b)COAMPS alongshore wind stress vs. distance to the coast at 3 different resolutions. The wind is averaged over a 30 km alongshore interval south of Pt. Sur (CalCS) during August 2003. From Capet et al. (2004).

Figure 1.17

the upwelling response. As an illustration of this sensitivity of the SST to the wind drop-off in the PCUS, the Figure 1.18 shows the Ekman advective heat flux for a strong CJ event from 29 to 30 October 2008, based on high resolution oceanic model simulations using contrasted drop-off patterns in surface wind forcing. Both simulations exhibit many mesoscale features related to eddies that induce positive and negative horizontal temperature gradients. The Ekman velocity is directed offshore along the coast and over the open ocean (not shown), which induces negative Ekman flux and SST cooling in the coastal region (50-km). The resulting offshore cooling transport is highly sensitive to the crosshore reduction of the momentum fluxes, and the sensitivity experiments indicate lower mean advective heat flux